Feasibility of the randomized response technique in rural Ethiopia.

A multiple answer model of Randomized Response Technique (known as Hopkins RRT Model II) was tried in a rural area (Nekempte) in Ethiopia to estimate the incidence of induced abortion among currently married women of childbearing age. In the RRT adopted here, the question on abortion--sensitive as it is--was preceded by two innocuous "practice questions". Despite the very low literacy level of the women, nearly all of them cooperated. The RRT estimates in regard to the two innocuous "practice questions" were fairly reasonable, while the RRT estimate of the rate of induced abortion (35 per cent) was far higher than that derived from direct reporting. The differentials in abortion rates by age and parity were consistent with expectation. A post RRT survey indicated that more than one-half (58 per cent) of the women found RRT "easy" or "moderately easy" to understand, while nearly 80 per cent of them thought that there was no "trick" involved in it.